Rework Integration Tests

[timuthy]

What you would like to be changed:
After having most of the etcd assets transitioned to components (see #278) we should revisit existing integration tests (envtests in /controllers) and reduce the size to an appropriate scope.

Motivation (Why is this needed?):
Today, etcd-druid has many integration tests that require a long time to run (~ 25m) and often test scenarios that are already covered by the lately introduced unit tests for different components.

For instance: It's redundant to have Envtests for every supported storage provider if we can verify the same in the scope of unit tests which are cheaper and more precise towards the test scenario, i.e. checking if the StatefulSet is created/updated properly.

A good summary about the purpose of integration tests can be found here: https://github.com/gardener/gardener/blob/master/docs/development/testing.md#purpose-of-integration-tests

At the same time, some of the existing integration tests do not properly clean up objects that were created before or during the tests. This potentially blocks controller workers for further tests and makes the entire suite flaky.

Approach/Hint to the implement solution (optional):
Please follow https://github.com/gardener/gardener/blob/master/docs/development/testing.md for best practices.

[abdasgupta]

We can optimize the tests in ETCD Druid by removing some of the redundant tests. We will pick up the task when we are less overloaded with other tasks.